Transcript
doi:10.1093/brain/aws282
Brain 2012: 135; 3770–3780
| 3770
BRAIN A JOURNAL OF NEUROLOGY
The degraded concept representation system in semantic dementia: damage to pan-modal hub, then visual spoke Paul Hoffman,1 Roy W. Jones2 and Matthew A. Lambon Ralph1 1 Neuroscience and Aphasia Research Unit (NARU), School of Psychological Sciences, University of Manchester, Manchester, M13 9PL, UK 2 Research Institute for the Care of Older People (RICE), Royal United Hospital, Bath, BA1 3NG, UK Correspondence to: Dr. Paul Hoffman, Neuroscience and Aphasia Research Unit (NARU), Zochonis Building, School of Psychological Sciences, University of Manchester, Oxford Road, Manchester, M13 9PL, UK E-mail:
[email protected]
The core clinical feature of semantic dementia is a progressive yet selective degradation of conceptual knowledge. Understanding the cognitive and neuroanatomical basis for this deficit is a key challenge for both clinical and basic science. Some researchers attribute the deficit to damage to pan-modal conceptual representations that are independent of any particular sensory–motor modality and are represented in the ventrolateral anterior temporal lobes. Others claim that damage to modality-specific visual feature representations in the occipitotemporal ‘ventral stream’ is responsible. In the present study, we tested the hypothesis that concept degradation in semantic dementia involves a combination of these pan-modal and modality-specific elements. We investigated factors influencing knowledge of object concepts by analysing 43 sets of picture-naming data from patients with semantic dementia. We found a strong influence of two pan-modal factors: highly familiar and typical items were named more accurately than less familiar/atypical items at all stages of the disorder. Items associated with rich sensory-motor information were also named more successfully at all stages, and this effect was present for sound/motion knowledge and tactile/action knowledge when these modalities were studied separately. However, there was no advantage for items rich in visual colour/form characteristics; instead, this factor had an increasingly negative impact in the later stages of the disorder. We propose that these results are best explained by a combination of (i) degradation of modality-independent conceptual representations, which is present throughout the disorder and is a consequence of atrophy focused on the ventrolateral anterior temporal lobes; and (ii) a later additional deficit for concepts that depend heavily on visual colour/form information, caused by the spreading of atrophy to posterior ventral temporal regions specialized for representing this information. This explanation is consistent with a graded hub-and-spoke model of conceptual knowledge, in which there is a gradual convergence of information along the temporal lobes, with visual attributes represented in the posterior cortex giving way to pan-modal representations in the anterior areas.
Keywords: conceptual knowledge; hub-and-spoke; ventral stream; anterior temporal lobe
Received April 18, 2012. Revised August 22, 2012. Accepted August 24, 2012 ß The Author (2012). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email:
[email protected]
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
Damaged concepts in semantic dementia
Introduction The goal of this study was to improve our understanding of the nature of conceptual knowledge impairment in semantic dementia, a neurodegenerative disorder in which conceptual knowledge progressively degrades, but other cognitive functions remain largely intact (Warrington, 1975; Hodges and Patterson, 2007). Semantic dementia is a highly selective disorder, at least in its early stages, both in terms of the cognitive deficit and the brain regions affected by atrophy. For this reason, semantic dementia has been influential in shaping theories of concept representation, making an understanding of this disorder important from both theoretical and clinical perspectives. In particular, this group of patients has been the inspiration behind what has become known as the hub-and-spoke theory of conceptual knowledge (Rogers et al., 2004a; Patterson et al., 2007; Pobric et al., 2010a). The ‘spokes’ in this model refer to a distributed network of brain regions that are specialized for representing conceptual information arising in different sensory–motor modalities. For example, the shape of a banana is thought to be coded in areas of the ventral temporal cortex specialized for representing the visual forms of objects (Martin et al., 1995; Chao et al., 1999; Thompson-Schill et al., 1999), the actions required to peel it in frontoparietal regions specialized for coding object manipulation (Buxbaum and Saffran, 2002; Kellenbach et al., 2003; Goldberg et al., 2006; Ishibashi et al., 2011) and its gustatory properties in orbitofrontal cortex (Goldberg et al., 2006). There is a broad consensus that such a network of modality-specific knowledge regions exists (Martin, 2007; Patterson et al., 2007; Barsalou, 2008; Binder and Desai, 2011; Kiefer and Pulvermuller, 2012). Damage to particular spoke regions has been proposed as an explanation for category-specific semantic disorders (Warrington and Shallice, 1984; Caramazza and Shelton, 1998; Capitani et al., 2003). For example, damage to inferior parietal regions involved in representing actions disproportionately affects processing associated with manipulation of tools and other manipulable man-made objects (Buxbaum and Saffran, 2002). The hub-and-spoke model also posits the existence of a pan-modal integrative ‘hub’ region, which receives inputs from the spoke regions and uses them to form transmodal conceptual representations that capture the deeper patterns of coherent variation across all sensory–motor and verbal modalities. These integrated representations are thought to be necessary because similarity in any single sensory–motor domain is, at best, only a partial guide to true conceptual similarity (Lambon Ralph et al., 2010). For example, apples and bananas have different colours and shapes but are conceptually similar. The proposal of a transmodal hub was motivated by the pan-modal pan-category conceptual deficits experienced by patients with semantic dementia. Patients with semantic dementia display poor conceptual knowledge in every modality tested, including written and spoken words (Jefferies et al., 2009), pictures (Bozeat et al., 2000), non-verbal sounds (Goll et al., 2010), smells (Luzzi et al., 2007), tastes (Piwnica-Worms et al., 2010) and object manipulation (Hodges et al., 2000). In addition, all categories of object are affected (Lambon Ralph et al., 2007), as are all types of words
Brain 2012: 135; 3770–3780
| 3771
(Hoffman and Lambon Ralph, 2011; Hoffman et al., 2012). These knowledge deficits are correlated with cortical atrophy and hypometabolism centred on the ventrolateral anterior temporal lobes (Nestor et al., 2006; Mion et al., 2010), suggesting that the pan-modal knowledge deficit is due to damage to a single neuroanatomically constrained region. This view is now supported by converging evidence from a number of other methodologies. Distortion-corrected functional MRI (Binney et al., 2010; Visser and Lambon Ralph, 2011), PET (Vandenberghe et al., 1996; Sharp et al., 2004), MEG (Marinkovic et al., 2003) and transcranial magnetic stimulation (Pobric et al., 2007, 2010b) have all revealed involvement of the anterior ventrolateral temporal lobes in processing various kinds of concepts presented in a number of different modalities. We refer to this perspective as the pan-modal view of semantic dementia because it emphasizes the involvement of verbal and non-verbal information from all sensory modalities in the conceptual breakdown. The hub-and-spoke model attributes this deficit to damage to a transmodal hub region that makes a critical contribution to all types of concept. In contrast, other researchers have highlighted the disproportionate loss of visual– perceptual knowledge in semantic dementia (Breedin et al., 1994; Bonner et al., 2009; Pulvermuller et al., 2010). Initial evidence for this assertion came from studies that probed knowledge for particular object features (Breedin et al., 1994; Tyler and Moss, 1998; Lambon Ralph et al., 1999, 2003). For example, when patients with semantic dementia are asked to define objects, their descriptions lack much of the information that healthy controls provide. Lambon Ralph et al. (2003) subdivided their responses according to the type of information provided (perceptual, functional or encyclopaedic) and found that patients with semantic dementia were more likely to omit the perceptual features of objects than their functional or encyclopaedic attributes. In addition, when patients are asked to draw common objects, their drawings lack many of the visual features produced by controls (Bozeat et al., 2003). These findings have led some to conclude that knowledge deficits in semantic dementia are primarily because of damage to areas of the ventral posterior temporal lobes specialized for coding the visual properties of objects (Yi et al., 2007; Bonner et al., 2010; Antonucci and Alt, 2011; Gainotti, 2011). This view is consistent with the observation that occasionally, individual patients with semantic dementia show better comprehension of abstract relative to concrete words (Warrington, 1975; Breedin et al., 1994; Yi et al., 2007; Bonner et al., 2009; Macoir, 2009; Papagno et al., 2009), although the reverse pattern is more common in case series of patients with semantic dementia (Jefferies et al., 2009; Hoffman and Lambon Ralph, 2011; Hoffman et al., 2012). The pan-modal and visual-deficit perspectives in semantic dementia are not mutually exclusive. Indeed, in terms of the hub-and-spoke model, it is possible that a combination of damage to the pan-modal hub region and the modality-specific visual spoke region gives rise to the pattern of deficits observed in semantic dementia (Bonner et al., 2009; Pulvermuller et al., 2010). To test this hypothesis, we investigated the degree to which the functions of the hub and spoke regions are impaired in semantic dementia, by analysing the patients’ performance in
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
3772
| Brain 2012: 135; 3770–3780
object naming. Object naming is a sensitive and reliable indicator of conceptual knowledge impairment in semantic dementia, and naming problems are evident even in the early stages of the disorder (Lambon Ralph et al., 2001; Woollams et al., 2008). We investigated the factors underpinning naming success in a large corpus of picture-naming data collected by our research group during the past 10 years (43 observations from 16 patients with semantic dementia). We assumed that the functional status of hub and spoke regions could be inferred from the patients’ success in naming objects with particular properties. The hub-and-spoke model states that all concepts rely on the transmodal hub, but, in addition, concepts rely on the individual modality-specific spokes to varying extents, depending on which modalities critically contribute to the concept (Pobric et al., 2010a). For example, in addition to the transmodal hub, the action spoke is thought to be heavily involved in naming ‘scissors’, as a key element of our knowledge of scissors is the way in which we manipulate them. In contrast, the action spoke is not expected to be important for naming an ‘elephant’; however, the visual spoke makes a critical contribution because an elephant’s most salient feature is its distinctive appearance. Thus, we assessed the status of the various spokes by contrasting sets of items with high versus low dependence on particular spoke regions (while controlling for other relevant factors). Greater impairment for items that rely heavily on a particular spoke region would indicate that the underlying representations in that spoke were compromised. To quantify the dependence of particular concepts on different spokes, we used data from a recent study in which participants rated the extent to which they associated objects with experience in eight sensory– motor modalities (colour, visual form, motion, sound, taste, smell, tactile and action; Hoffman and Lambon Ralph, submitted for publication). This allowed us to determine (i) whether objects that rely heavily on visual information are particularly impaired in the disorder, as predicted by the visual deficit hypothesis; and (ii) whether information in other sensory–motor modalities was similarly affected. We also assessed the impact of three pan-modal object properties that are not tied to any particular sensory modality: familiarity, typicality and sensory richness. If the loss of concepts in semantic dementia is primarily because of degradation of visual feature knowledge, these factors would not be especially influential because none is based directly on information from this modality. In contrast, the pan-modal hub theory of semantic dementia holds that these factors have a strong influence on success across a range of semantic tasks, including naming (Funnell, 1995; Patterson et al., 2007; Woollams et al., 2008; Jefferies et al., 2009; Mayberry et al., 2011). Highly familiar and typical items are thought to be represented more strongly in the hub, making them more robust to damage (Rogers et al., 2004a). As a new additional measure of the transmodal hub contribution to semantic representation, we assessed whether knowledge in semantic dementia was influenced by the total amount of sensory–motor information associated with the concept over all modalities, which we termed its ‘sensory richness’.
P. Hoffman et al.
Materials and methods Participants This study features data from 16 patients with semantic dementia, collected between 2001 and 2010. The patients were referred from the Research Institute for the Care of Older People memory clinic in Bath, UK, or from other specialist clinics in the north-west of England. Each had received a clinical diagnosis of semantic dementia based on caregiver interview, neuropsychological assessment and evidence of temporal lobe atrophy in structural brain imaging. All patients fulfilled current diagnostic criteria for semantic dementia (Gorno-Tempini et al., 2011). Most have appeared in one or more previous studies by our research group (for example Jefferies et al., 2004; Patterson et al., 2006; Lambon Ralph et al., 2010; Hoffman and Lambon Ralph, 2011). Our analyses were conducted on data from the 64-item picturenaming test that forms part of the Cambridge Semantic Battery (Bozeat et al., 2000), which we use routinely to assess the integrity of conceptual knowledge. As patients were studied longitudinally, many completed this test on multiple occasions. Because of the progressive nature of semantic dementia, and in keeping with previous studies (Lambon Ralph et al., 2001; Woollams et al., 2007, 2008), we treated each test administration as an independent observation. The delay between successive administrations of the naming test was typically between 6 and 12 months and was never 53 months. The 16 patients yielded a total of 43 observations. These were divided into three groups by classifying the 14 highest-naming scores as ‘mild’, the 14 lowest scores as ‘severe’ and the remaining 15 scores as ‘moderate’.
Background neuropsychological testing Data from standard neuropsychological tests are presented in Table 1, which shows mean scores for the mild, moderate and severe naming groups. In addition, data from each individual patient, taken from the first occasion on which they completed the full assessment battery, are provided in the Supplementary material. In the most recent incarnation of our assessment regime, patients complete a full assessment battery once a year but additionally complete core semantic knowledge tests (i.e. picture naming and word–picture matching) every 6 months. For this reason, there are some occasions on which a naming score was available, but other tests were not completed. All three groups performed outside the normal range on a series of tests that probes semantic/conceptual knowledge. These consisted of picture naming, matching spoken words to pictures, the Camel and Cactus Test (a semantic association test similar to the Pyramids and Palm Trees Test; see Bozeat et al., 2000) and generation of examples belonging to six semantic categories (category fluency). Naming was considerably impaired even in the mild group (the cut-off for normal performance is 59/64 on this relatively easy test) and declined steadily. On the other tests, performance declined in parallel with naming, indicating that the naming deficit in these patients was symptomatic of a more general deterioration in conceptual knowledge. In contrast, other cognitive functions were relatively spared, in keeping with the usual clinical picture in semantic dementia. General cognitive status was assessed with the Mini-Mental State Examination (Folstein et al., 1975). Attention and executive function were tested using forward and backward digit span (Wechsler, 1987), letter fluency (FAS) and Raven’s coloured progressive matrices (Raven, 1962). Finally, visuospatial function was assessed with copying of the Rey Complex Figure (Lezak, 1976) and four subtests from the Visual
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
Damaged concepts in semantic dementia
Brain 2012: 135; 3770–3780
| 3773
Table 1 Mean test scores for each group Test/factor
Age (years) Education (leaving age, years) Picture naming Category fluency (six categories) Word–picture matching Camel and Cactus Test (pictures) Letter fluency (FAS) MMSE Digit span forwards Digit span backwards Raven’s progressive matrices Rey figure copy VOSP: incomplete letters VOSP: object decision VOSP: number location VOSP: cube analysis
Max
64 64 64 30
36 36 20 20 10 10
Mild
Moderate
Severe
Mean
n
Mean
n
Mean
n
62 15.8 43 48 55 45 19 27 6.5 4.8 33 35 19.3 17.0 9.7 9.6
14 14 14 7 12 8 7 7 8 8 7 8 7 7 7 7
64 15.5 22 26 41 31 21 25 6.3 4.3 28 29 16.9 13.6 9.5 9.3
15 15 14 11 13 10 11 9 12 12 11 12 11 11 11 11
68 16.3 9 17 31 32 17 21 5.3 4.4 30 33 15.0 13.9 9.5 8.5
14 14 14 9 13 9 9 8 10 9 10 10 10 10 10 10
Healthy controls Mean (cut-off)
62.3 95.7 63.8 59.1 41.1 28.8 6.8 4.8
(59) (62) (62) (51) (17) (24) (4) (3)
34.0 18.8 17.7 9.4 9.2
(31) (16) (14) (7) (6)
Impaired scores are shown in bold (below published cut-offs 42 SDs below the mean in healthy subjects). Numbers (n) indicate the number of observations contributing to each mean. MMSE = Mini-Mental State Examination; VOSP = visual object and space perception battery.
Object and Space Perception Battery (Warrington and James, 1991). The mild group displayed no impairments on any of these tests. There was evidence for some slight decline in the moderate and severe groups, most notably on the Mini-Mental State Examination and the object decision component of the Visual Object and Space Perception Battery. These can be attributed to deterioration in language comprehension/production, which is an integral part of the Mini-Mental State Examination, and in object recognition, which declines as patients lose their knowledge of previously familiar objects (Rogers et al., 2004b). A mild impairment in letter fluency was evident in the most severe group, reflecting their impoverished vocabulary.
Stimuli and test administration The Cambridge picture-naming test consists of 64 black and white line drawings from the Snodgrass and Vanderwart (1980) set. Half represent living and half non-living objects. During the test, each picture was presented individually to the patient, and they were asked to give its name. The item was scored as correct if they provided its name, or any of the acceptable alternatives listed by Woollams et al. (2008), as part of their response.
Stimulus factors and data analysis We focused on six factors of potential importance to object knowledge in semantic dementia. With the exception of typicality, all factors were based on ratings collected by Hoffman and Lambon Ralph (submitted for publication) from 100 British undergraduate students. Participants in this study were presented with a series of object names (animals, fruits and vegetables and manmade objects) and were asked to rate, on a 7-point scale, (i) how familiar they were with the object; and (ii) how strongly they associated the object with experience in each of the eight sensory–motor modalities (colour, visual form, observed motion, sound, taste, smell, tactile and performed actions). We also used ratings of semantic typicality collected by Garrard et al. (2001). In this study, participants were presented with object names and asked
to rate, on a 4-point scale, whether the object was a typical example of items from its superordinate category (e.g. a dog is a highly typical animal and a seahorse atypical). From this pool of data, we derived six factors. The first two factors, ‘familiarity’ and ‘typicality’, were pan-modal in the sense that raters were not asked to take any particular modalities into account when making their judgements. We assumed, therefore, that raters made use of all types of information available for the concept when making these ratings. The third factor, ‘sensory richness’, was a composite measure of the total amount of sensory–motor experience associated with each object. It was calculated by taking the average of each item’s experience ratings across all of the eight modalities. The remaining three factors indexed how strongly each item was associated with experiences in particular sensory–motor domains. We created three composite measures by combining ratings from closely related modalities. The decision to group modalities in this way was guided by a previous study in which we performed a principal components analysis on the ratings for 156 objects (Hoffman and Lambon Ralph, submitted for publication). This analysis indicated that the eight modalities collapsed into four distinct factors, with two modalities loading heavily on each factor, Factor 1: taste and smell, Factor 2: sound and motion, Factor 3: colour and visual form and Factor 4: tactile and action. These groupings capture a number of intuitive assumptions about modality correlations. For example, items with distinctive smells can often be eaten, and objects that move (principally animals and vehicles) usually make sounds. For the present study, we discarded the taste and smell factor because these modalities are irrelevant for the majority of objects. The other modalities were used to construct the final three factors in the following manner: (i) ‘colour/form’ value of the object, calculated by averaging its ratings for colour experience and visual form experience; (ii) ‘sound/motion’ value of the object, calculated by averaging its ratings for sound experience and experience of observed motion; and (iii) ‘tactile/action’ value of the object, calculated by averaging its ratings for tactile experience and action experience.
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
3774
| Brain 2012: 135; 3770–3780
This approach for grouping modalities also aligns with the presumed neuroanatomical correlates of each modality. Colour and visual form information are both thought to be underpinned by posterior fusiform areas that form part of the ventral visual stream (Martin et al., 1995; Chao et al., 1999; Thompson-Schill et al., 1999). Motion characteristics are associated with a different locus of activation in the posterior middle temporal gyrus (Beauchamp et al., 2002), which is close to the superior temporal region activated during processing of the auditory properties of objects (Lewis et al., 2004; Kiefer et al., 2008). Finally, tactile information and knowledge of the actions associated with objects are thought to be stored in more dorsal parietal and frontal sites (Buxbaum and Saffran, 2002; Kellenbach et al., 2003; Goldberg et al., 2006; Ishibashi et al., 2011). To assess the effect of each of these factors on conceptual knowledge in semantic dementia, we constructed 12 subsets of 16 items, all taken from the pool of 64 items in the naming test. The subsets were arranged in six pairs, with each pair manipulating one of the factors while being matched as closely as possible on the other five. The properties of each subset are shown in Table 2. Note that each pair differed significantly on exactly one of the six factors. In addition, none of the pairs differed in log word frequency (taken from the CELEX database; Baayen et al., 1993), and each subset contained an equal number of living and non-living things. The items in each subset are listed in the Supplementary material.
Statistical analysis Data from each stimulus manipulation were analysed separately. Results from each manipulation were subjected to a 2 3 mixed ANOVA that included the manipulated variable as a within-subjects factor and severity as a between-subjects factor. As expected, the main effect of severity was highly significant in every analysis. Therefore, we report only main effects of the manipulated stimulus factors and their interactions with severity. Post hoc t-tests were used to explore the significance of effects at each level of severity.
Results Stimulus factors Familiarity Naming performance for high- and low-familiarity items is shown in Fig. 1. Highly familiar items were named much more successfully than less familiar items, which was confirmed by the significant main effect of familiarity in the 2 3 ANOVA [F(1,40) = 105, P 5 0.001]. There was also a significant interaction between familiarity and severity [F(2,40) = 8.09, P = 0.001]. t-Tests indicated that the familiarity effect was highly significant at every level of severity (t 4 4.42, P 5 0.002). However, it was smallest in the severe group, presumably reflecting a slight floor effect in the most severe patients.
Typicality Results for the high- and low-typicality sets are shown in Fig. 1. Highly typical items were more likely to be named than less typical ones [F(1,40) = 31.5, P 5 0.001], and this effect did not interact with severity [F(2,40), P 5 1). The effect was highly significant in every severity group (t 4 3.11, P 5 0.009).
P. Hoffman et al.
Sensory richness Results for the overall richness across all sensory modalities are presented Fig. 1. Patients were more accurate at naming objects with high sensory richness [F(1,40) = 31.5, P 5 0.001]. This factor did not interact with severity [F(2,40), P 5 1] and was highly significant at every stage of the disease (t 4 4.41, P 5 0.002).
Colour/form Figure 2 shows the manipulations of representation in particular sensory modalities. In contrast to the results for ‘overall’ sensory richness, items that had strong colour and form characteristics were named more poorly than matched items with weaker colour/form ratings [F(1,40) = 16.8, P 5 0.001]. However, this effect was qualified by a significant interaction [F(2,40) = 4.18, P = 0.022]. t-Tests indicated that colour/form had no effect on naming in the mildest affected patients (t = 0.18, P = 0.86), a modest effect on the moderately affected patients (t = 2.58, P = 0.022) and the strongest effect on the most severe patients (t = 4.60, P 5 0.001).
Sound/motion The effect of the sound/motion manipulation (Fig. 2) mirrored that of overall sensory richness. Items that were high in sound/ motion information were named more successfully [F(1,40) = 91.3, P 5 0.001]. There was no interaction with severity [F(2,40) P 5 1], and the effect was highly significant in all three groups (t 4 3.57, P 5 0.004).
Tactile/action Finally, the manipulation of tactile/action qualities is shown in Fig. 2. The effect of this manipulation was significant [F(1,40) = 18.7, P 5 0.001], indicating a naming advantage for items rated strongly for tactile/action knowledge. There was no interaction with severity [F(2,40) = 1.04, P = 0.36]; however, when the three groups were considered separately, the effect only reached significance in the moderate (t = 2.28, P = 0.038) and severe (t = 5.00, P 5 0.001) groups.
Are high colour/form items more difficult to name? Analysis of picture-naming results indicated that highly familiar and typical items were more likely to be named correctly at all stages of the disorder and that high sensory richness also benefited naming. This richness advantage was also apparent for sound/motion and tactile/action modalities when these domains were studied separately. Conversely, no advantage was observed for items with rich colour/form characteristics; in fact, richness in this domain had a negative effect on performance in the more severe patients. This result supports the idea that representations of visual attributes are particularly affected later in the disorder. However, another possibility is that high colour/form items were simply more difficult to name, perhaps as colour plays an important role in their identification, and this information was absent from the black-and-white drawings used. To rule out this possibility, we turned to reaction time data collected from 17 healthy subjects (mean age = 22 years) who named the same line drawings. Pictures were presented on a 15-inch computer monitor, and
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
Damaged concepts in semantic dementia
Brain 2012: 135; 3770–3780
| 3775
Figure 1 Manipulations of familiarity, typicality and sensory richness. High and low conditions differ at **P 5 0.01. Bars indicate 1 SEM.
Figure 2 Manipulations of richness in particular sensory modalities. High and low conditions differ at *P 5 0.05 and **P 5 0.01.
participants were asked to name each picture as quickly as possible. No time limit was placed on responses. Reaction times were recorded using an electronic voice key, with responses scored online and digitally recorded for later verification. If high colour/ form items were more difficult to name, one would expect these items to be named more slowly by healthy participants. In fact, the opposite was observed. Reaction times for high-colour/form items (mean = 848 ms) were considerably faster than those for low-colour/form items (mean = 991 ms), and this difference was highly significant [t(16) = 6.73, P 5 0.001]. Error rates were low and did not differ between the two conditions. This result confirms that high-colour/form items are intrinsically easier to name in the healthy population, making the significant disadvantage for these items in the cases with severe semantic dementia all the more striking. Indeed, the reaction time advantage for high-colour/ form items in healthy subjects contrasts with the lack of any such advantage in the mild semantic dementia group, suggesting that some subtle impairment to visual attributes may be present even at an early stage.
Discussion Though there is no doubt that patients with semantic dementia present with selective conceptual knowledge deficits, the
underlying cause of this impairment remains a subject of active debate. Some researchers hold that the deficit primarily reflects damage to a pan-modal hub, which develops representations that are independent of any particular sensory modality and license the formation of coherent concepts (Patterson et al., 2007; Lambon Ralph et al., 2010). Others propose that deficits in semantic dementia are primarily caused by disruption to areas of the cortex that are specifically implicated in representing the visual characteristics of objects (Breedin et al., 1994; Bonner et al., 2009; Macoir, 2009). We investigated these two possibilities by exploring the impact of pan-modal and modality-specific stimulus factors on conceptual knowledge in semantic dementia. Naming was strongly influenced by pan-modal factors. Patients were more likely to name items that were highly familiar and highly typical of their class. We also found an advantage for concepts with high overall sensory richness, which was also observed in two specific modalities. Items were more likely to be named correctly if they were rich in sound/motion or tactile/action characteristics. In contrast to all the other factors, a different pattern emerged for items rich in visual colour/form information. There was no advantage for these concepts and, in the more severe patients, such items were instead disproportionately impaired. How are these findings best reconciled with the theories of semantic dementia described earlier? We propose that the full
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
3776
| Brain 2012: 135; 3770–3780
pattern of results is best explained through a combination of continually increasing atrophy to pan-modal representations and later damage to visual-specific areas of cortex. Specifically, we suggest that the initial cause of knowledge impairment is damage to the pan-modal representations stored in the ventrolateral anterior temporal lobes. This remains the primary cause of impairment throughout the progression of the disease, as atrophy remains centred on the same anterior region, explaining the large effects of pan-modal factors at all stages of severity. As the disease becomes more severe, however, additional impairments begin to emerge for concepts that rely heavily on visual characteristics. These are most likely to reflect the later spread of atrophy, which begins to envelop the posterior ventral temporal lobes (Bright et al., 2008; Rohrer et al., 2009), which are specialized for processing the visual features of objects (Martin et al., 1995; Chao et al., 1999; Thompson-Schill et al., 1999). We will consider the evidence for pan-modal degradation first. As discussed earlier, conceptual knowledge in semantic dementia is lost for information presented in all sensory modalities and for all categories of object (Bozeat et al., 2000; Patterson et al., 2007). This pattern is unlikely to occur as a result of damage to a single modality-specific system. Instead, the hub-and-spoke theory holds that atrophy centred on the ventrolateral anterior temporal lobes affects all types of concepts because it damages central modality-invariant representations (Patterson et al., 2007; Lambon Ralph et al., 2010; Pobric et al., 2010a). These representations are learned through the co-activation of information in different sensory–motor modalities but are themselves independent of any particular modality (Rogers et al., 2004a). Damage to these representations consequently affects all concepts, regardless of which particular modalities they are associated with. Neuroimaging evidence indicates that these representations are primarily underpinned by the ventrolateral anterior temporal lobes (centred on the anterior fusiform and inferior temporal gyri). This region is activated by a range of semantic processing tasks, including understanding of spoken words (Sharp et al., 2004); comprehension of concrete and abstract written words (Binney et al., 2010); classification of objects presented as words, pictures and environmental sounds (Visser and Lambon Ralph, 2011) and judgements of semantic association for word and picture stimuli (Visser et al., 2012). In line with this evidence, a recent study of patients with semantic dementia found that the degree of hypometabolism in the anterior fusiform correlated with their performance on tests of conceptual knowledge (Mion et al., 2010). Hypometabolism in other regions was not related to semantic performance, even though the entire anterior temporal lobe, including the pole, was markedly hypometabolic. This study, taken with the converging evidence from functional neuroimaging in neurologically intact participants, helps to pinpoint the anterior fusiform as the centre point of the pan-modal ‘hub’ cortex. Consistent with the hub-and-spoke theory, in the present study, we found that two pan-modal stimulus factors, familiarity and typicality, had strong effects on naming in semantic dementia, at all stages of disease severity. These effects have been reported previously in patients with semantic dementia (Lambon Ralph et al., 1998; Woollams et al., 2008) and are consistent with
P. Hoffman et al. breakdown in modality-invariant conceptual knowledge. The representations of highly familiar concepts are more strongly instantiated because the system has more opportunities to acquire their meanings. They also tend to be acquired earlier in life and to be encountered more often during the disease process, which may contribute to their relative preservation in the face of the progressive representational degradation (Lambon Ralph et al., 1998; Jefferies et al., 2011). This phenomenon has been formally simulated in computational models that use a layer of modalityindependent processing units to capture conceptual structure. These simulations indicate that more familiar concepts come to occupy a larger portion of the semantic space formed by the modality-independent layer, providing them with a greater robustness to damage (Simulation 5.4; Rogers and McClelland, 2004). Highly typical items may be preserved for similar reasons—typical exemplars, by definition, are similar in meaning and share many attributes in common. This means that, during development, learning about one typical exemplar will also boost the representations of other overlapping concepts within that semantic cohort. In comparison, atypical concepts do not overlap with as many other items and thus do not benefit from this same type of collateral boost. We found that overall sensory richness, based on an item’s average ratings of association with all sensory–motor modalities, was positively related to the level of remaining item knowledge in the cases with semantic dementia. Concepts that were associated with a rich set of sensory–motor information were more likely to be named correctly, even though they were matched to less rich concepts in terms of familiarity and typicality. This positive effect supports the idea that rich sources of information about a concept lead to a strong representation that is relatively robust to damage (Plaut and Shallice, 1993). Words associated with richer semantic representations are also processed more efficiently by healthy subjects in a range of tasks (Pexman et al., 2008). Similarly, we found that, within the domain of concrete objects, weaker sensory– motor information was associated with poorer object knowledge at all stages of semantic dementia. Extrapolating from this trend, one would expect abstract concepts, associated with more impoverished sensory–motor information, to be even more susceptible to damage. Case series investigations suggest that this is indeed the case for the majority of patients with semantic dementia (Jefferies et al., 2009; Hoffman and Lambon Ralph, 2011). In a small number of cases, the reverse pattern has been observed (e.g. Warrington, 1975; Breedin et al., 1994; Macoir, 2009), though these effects may be because of individual differences in the location of cortical atrophy or in premorbid experience (Hoffman and Lambon Ralph, 2011; Jefferies et al., 2011). In contrast to the positive overall effect of rich sensory–motor information, there was no advantage associated with strong visual colour/form characteristics. Instead, items with strong colour/form ratings were named more poorly than matched items for which such visual information was less prominent. This effect was not present in the mildest cases but was apparent, with increasing magnitude, in the moderate and severe cases. This finding fits with previous investigations, which found that, within the context of generally impoverished feature knowledge, patients with semantic dementia were especially poor at reporting visual attributes
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
Damaged concepts in semantic dementia
Brain 2012: 135; 3770–3780
| 3777
Table 2 Mean (standard deviation) properties of each stimulus set Manipulation
Item set
Familiarity
Typicality
Sensory richness
Colour and form
Sound and motion
Tactile and action
Log word frequency
Familiarity
High Low High Low High Low High Low High Low High Low
6.52 5.27 6.04 5.96 6.11 5.89 5.94 5.84 5.90 6.03 6.15 6.02
3.10 2.96 3.55 2.13 2.97 2.91 3.06 3.06 2.88 2.83 3.03 2.89
4.08 3.85 4.00 3.95 4.33 3.62 4.03 3.77 4.04 3.78 4.12 3.91
5.73 5.48 5.54 5.84 5.80 5.65 5.94 5.25 5.51 5.70 5.57 5.63
3.95 4.79 4.49 3.76 4.62 3.90 4.40 4.53 5.32 3.07 4.40 4.38
4.07 3.80 3.88 3.92 4.13 3.68 3.84 3.79 3.75 3.96 4.55 3.53
1.34 1.14 1.07 1.00 1.34 1.21 1.16 1.18 1.15 1.11 1.29 1.09
Typicality Sensory richness Colour/form Sound/motion Tactile/action
(0.33) (0.45) (0.75) (0.56) (0.58) (0.78) (0.77) (0.80) (0.74) (0.77) (0.75) (0.59)
(0.57) (0.76) (0.23) (0.67) (0.88) (0.84) (0.73) (0.73) (1.0) (0.87) (0.78) (0.81)
(0.58) (0.38) (0.48) (0.53) (0.39) (0.31) (0.51) (0.44) (0.41) (0.68) (0.36) (0.61)
(0.78) (0.55) (0.74) (0.75) (0.76) (0.67) (0.59) (0.53) (0.71) (0.72) (0.65) (0.79)
(1.49) (1.19) (1.38) (1.35) (1.59) (1.01) (1.04) (1.26) (0.70) (0.64) (1.23) (1.32)
(0.66) (0.83) (0.81) (0.84) (0.67) (0.89) (0.80) (0.81) (0.89) (0.84) (0.63) (0.65)
(0.38) (0.43) (0.41) (0.41) (0.46) (0.33) (0.51) (0.47) (0.43) (0.44) (0.59) (0.40)
High and low values shown in bold differed significantly from each other (P 5 0.005). No other pairs differed significantly (P 4 0.09).
(Breedin et al., 1994; Tyler and Moss, 1998; Lambon Ralph et al., 1999, 2003). Before turning to our preferred interpretation of these findings, there are a couple of alternative possibilities to address. The representations of living things are often thought to be more strongly dependent on visual information than those of man-made objects (Warrington and Shallice, 1984; Farah and McClelland, 1991), raising the possibility of a semantic category effect underpinning the observed reversed effect of visual colour richness. However, the present results cannot be interpreted as a category effect because our high- and low-colour/form sets contained equal numbers of living and non-living items. Indeed, dissociations between living and non-living items are rare in semantic dementia (Lambon Ralph et al., 2007). Nor was it the case that high-colour/form items were intrinsically more difficult to name, as healthy subjects showed a processing advantage for these items. Instead, these results suggest that objects that are strongly associated with visual information are disproportionately impaired in the more severe stages of semantic dementia, in line with damage to visual association cortex. This view is also consistent with the known progression of cerebral atrophy in semantic dementia. Representations of colour and visual form are associated with the posterior ventral temporal lobe, particularly the posterior fusiform gyrus (Chao et al., 1999, Martin, 2007). Importantly, this region appears to be involved in conceptual knowledge of visual object properties, and not merely processing of visual input, as it is activated when visual property knowledge is probed verbally (Thompson-Schill et al., 1999; Kellenbach et al., 2001). The ventral posterior temporal lobe is typically spared in the early stages of semantic dementia, in which the damage principally affects the temporal poles and the ventrolateral surface of the anterior temporal lobes (evident in terms of atrophy, hypometabolism and reduced connectivity; Galton et al., 2001; Nestor et al., 2006; Acosta-Cabronero et al., 2011; Galantucci et al., 2011). As the disease progresses, atrophy in these regions increases. In addition, cortical thickness measures and voxel-based morphometry indicate that the increasing atrophy begins to envelop posterior ventral temporal regions (Bright et al., 2008; Rohrer et al., 2009). This posterior spread of atrophy could be expected to disproportionately affect concepts that rely heavily on the visual attribute information assumed to be
coded by the region. In contrast, the posterior middle and superior temporal gyri, associated with sound and motion characteristics, are less severely affected as atrophy spreads, as are the parietal and frontal sites associated with tactile information and knowledge of object manipulation. In the present study, we probed conceptual knowledge through picture naming because this task is reliably impaired at all stages of the disorder and because it is clear that anomia in semantic dementia is the consequence of core conceptual deterioration and not a visually specific deficit. However, the use of pictures raises an important question: is the visual spoke region solely a representational area, concerned with representing object properties that are usually experienced through vision, or is it directly involved in the perceptual identification of visually presented objects? This distinction has sometimes caused confusion in discussion of modality-specific semantic systems (Coltheart et al., 1998). In our view, however, these two positions are difficult to disentangle. Identification of objects from vision relies heavily on having intact semantic knowledge of how objects usually appear. Similarly, identifying an object from its sound depends critically on an intact representation of the object’s auditory properties. Therefore, if we assume that the visual spoke region codes the typical visual characteristics of ‘giraffes’, for example, then it follows that this region (in concert with the hub) will make a critical contribution when someone is asked to identify a picture of a giraffe. The important point from our perspective is that spoke regions contribute to conceptual knowledge even when the modality of input does not match the modality of the probed knowledge. Posterior ventral fusiform activation has been demonstrated, for example, when visual object properties are probed verbally (Thompson-Schill et al., 1999; Kellenbach et al., 2001), inferior parietal regions associated with action knowledge are activated when object manipulation judgements are made in response to static pictures (Van Dam et al., 2012) and superior temporal regions linked to auditory knowledge are activated when the names of sound-making objects are presented as written words (Kiefer et al., 2008). It seems likely, therefore, that each spoke region contributes to semantic processing whenever attributes in its modality are relevant to the task in hand, irrespective of the modality of the sensory input.
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
3778
| Brain 2012: 135; 3770–3780
To summarize, our investigation of the factors underpinning impairment of object knowledge in semantic dementia supports a multi-component model, which is consistent with the known progression of cortical atrophy in the disorder. Based on the hub-and-spoke theory of conceptual knowledge, we propose that atrophy to the ventrolateral anterior temporal lobes, present from the earliest stages of neurodegeneration, affects transmodal ‘hub’ representations that are independent of any particular sensory–motor modality. This results in a pan-modal pan-categorical knowledge deficit that remains the principal feature throughout the disorder as the severity of atrophy in this region increases. Later in the disorder, the boundary of atrophy extends into the posterior ventral temporal lobes, affecting a modality-specific ‘spoke’ region that codes the visual attributes of objects. This later damage begins to disproportionately affect concepts that rely heavily on this source of information. By positing a multi-component explanation, this approach has the potential to reconcile the differing viewpoints on the functional basis of conceptual impairment in semantic dementia.
Acknowledgements The authors thank Beth Jefferies and Hannah Gardner for their help with data collation, and to Alistair Burns and Roland Zahn for referring patients to us.
Funding The research was supported by two Medical Research Council programme grants (G0501632 and MR/J004146/1) and an senior investigator grant awarded to M.A.L.R. by the UK National Institute for Health Research.
Supplementary material Supplementary material is available at Brain online.
References Acosta-Cabronero J, Patterson K, Fryer TD, Hodges JR, Pengas G, Williams GB, et al. Atrophy, hypometabolism and white matter abnormalities in semantic dementia tell a coherent story. Brain 2011; 134: 2025–35. Antonucci SM, Alt M. A lifespan perspective on semantic processing of concrete concepts: does a sensory/motor model have the potential to bridge the gap? Cogn Affect Behav Neurosci 2011; 11: 551–71. Baayen RH, Piepenbrock R, van Rijn H. The CELEX lexical database (CD-ROM). Philadelphia, PA: Linguistic Data Consortium, University of Pennsylvania; 1993. Barsalou LW. Grounded cognition. Annu Rev Psychol 2008; 59: 617–45. Beauchamp MS, Lee KE, Haxby JV, Martin A. Parallel visual motion processing streams for manipulable objects and human movements. Neuron 2002; 34: 149–59. Binder JR, Desai RH. The neurobiology of semantic memory. Trends Cogn Sci 2011; 15: 527–36. Binney RJ, Embleton KV, Jefferies E, Parker GJM, Lambon Ralph MA. The inferolateral aspects of the anterior temporal lobe are crucial in semantic memory: evidence from a novel direct comparison of
P. Hoffman et al. distortion-corrected fMRI, rTMS and semantic dementia. Cereb Cortex 2010; 20: 2728–38. Bonner MF, Ash S, Grossman M. The new classification of primary progressive aphasia in semantic, logopenic, or nonfluent/agrammatic variants. Curr Neurol Neurosci Rep 2010; 10: 484–90. Bonner MF, Vesely L, Price C, Anderson C, Richmond L, Farag C, et al. Reversal of the concreteness effect in semantic dementia. Cogn Neuropsychol 2009; 26: 568–79. Bozeat S, Lambon Ralph MA, Graham KS, Patterson K, Wilkin H, Rowland J, et al. A duck with four legs: investigating the structure of conceptual knowledge using picture drawing in semantic dementia. Cogn Neuropsychol 2003; 20: 27–47. Bozeat S, Lambon Ralph MA, Patterson K, Garrard P, Hodges JR. Non-verbal semantic impairment in semantic dementia. Neuropsychologia 2000; 38: 1207–15. Breedin SD, Saffran EM, Coslett HB. Reversal of the concreteness effect in a patient with semantic dementia. Cogn Neuropsychol 1994; 11: 617–60. Bright P, Moss HE, Stamatakis EA, Tyler LK. Longitudinal studies of semantic dementia: the relationship between structural and functional changes over time. Neuropsychologia 2008; 46: 2177–88. Buxbaum LJ, Saffran EM. Knowledge of object manipulation and object function: dissociations in apraxic and nonapraxic subjects. Brain Lang 2002; 82: 179–99. Capitani E, Laiacona M, Mahon B, Caramazza A. What are the facts of semantic category-specific deficits? A critical review of the clinical evidence. Cogn Neuropsychol 2003; 20: 213–61. Caramazza A, Shelton JR. Domain-specific knowledge systems in the brain: the animate-inanimate distinction. J Cogn Neurosci 1998; 10: 1–34. Chao LL, Haxby JV, Martin A. Attribute-based neural substrates in temporal cortex for perceiving and knowing about objects. Nat Neurosci 1999; 2: 913–9. Coltheart M, Inglis L, Cupples L, Michie P, Bates A, Budd B. A semantic subsystem of visual attributes. Neurocase 1998; 4: 353–70. Farah MJ, McClelland JL. A Computational model of semantic memory impairment: modality specificity and emergent category specificity. J Exp Psychol Gen 1991; 120: 339–57. Folstein MF, Folstein SE, McHugh PR. Mini-mental state: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12: 189–98. Funnell E. Objects and properties: a study of the breakdown of semantic memory. Memory 1995; 3: 497–518. Gainotti G. The organization and dissolution of semantic-conceptual knowledge: is the ‘amodal hub’ the only plausible model? Brain Cogn 2011; 75: 299–309. Galantucci S, Tartaglia MC, Wilson SM, Henry ML, Filippi M, Agosta F, et al. White matter damage in primary progressive aphasias: a diffusion tensor tractography study. Brain 2011; 134: 3011–29. Galton CJ, Patterson K, Graham K, Lambon-Ralph MA, Williams G, Antoun N, et al. Differing patterns of temporal atrophy in Alzheimer’s disease and semantic dementia. Neurology 2001; 57: 216–25. Garrard P, Lambon Ralph MA, Hodges JR, Patterson K. Prototypicality, distinctiveness, and intercorrelation: analyses of the semantic attributes of living and nonliving concepts. Cogn Neuropsychol 2001; 18: 125–74. Goldberg RF, Perfetti CA, Schneider W. Perceptual knowledge retrieval activates sensory brain regions. J Neurosci 2006; 26: 4917–21. Goll JC, Crutch SJ, Loo JHY, Rohrer JD, Frost C, Bamiou DE, et al. Non-verbal sound processing in the primary progressive aphasias. Brain 2010; 133: 272–85. Gorno-Tempini ML, Hillis AE, Weintraub S, Kertesz A, Mendez M, Cappa SF, et al. Classification of primary progressive aphasia and its variants. Neurology 2011; 76: 1006–14. Hodges JR, Bozeat S, Lambon Ralph MA, Patterson K, Spatt J. The role of conceptual knowledge in object use: evidence from semantic dementia. Brain 2000; 123: 1913–25.
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
Damaged concepts in semantic dementia Hodges JR, Patterson K. Semantic dementia: a unique clinicopathological syndrome. Lancet Neurol 2007; 6: 1004–14. Hoffman P, Jones RW, Lambon Ralph MA. Be concrete to be comprehended: consistent imageability effects in semantic dementia for nouns, verbs, synonyms and associates. Cortex 2012. Advance Access published on May 26, 2012, doi: 10.1016/j.cortex.2012.05.007. Hoffman P, Lambon Ralph MA. Reverse concreteness effects are not a typical feature of semantic dementia: evidence for the hub-and-spoke model of conceptual representation. Cereb Cortex 2011; 21: 2103–12. Hoffman P, Lambon Ralph MA. Shapes, scents and sounds: quantifying the full multi-sensory basis of conceptual knowledge. Submitted for publication. Ishibashi R, Lambon Ralph MA, Saito S, Pobric G. Different roles of lateral anterior temporal lobe and inferior parietal lobule in coding function and manipulation tool knowledge: evidence from an rTMS study. Neuropsychologia 2011; 49: 1128–35. Jefferies E, Patterson K, Jones RW, Bateman D, Lambon Ralph MA. A category-specific advantage for numbers in verbal short-term memory: evidence from semantic dementia. Neuropsychologia 2004; 42: 639–60. Jefferies E, Patterson K, Jones RW, Lambon Ralph MA. Comprehension of concrete and abstract words in semantic dementia. Neuropsychology 2009; 23: 492–9. Jefferies E, Rogers TT, Lambon Ralph MA. Premorbid expertise produces category-specific impairment in a domain-general semantic disorder. Neuropsychologia 2011; 49: 3213–23. Kellenbach M, Brett M, Patterson K. Large, colourful or noisy? Attributeand modality-specific activations during retrieval of perceptual attribute knowledge. Cogn Affect Behav Neurosci 2001; 1: 207–21. Kellenbach ML, Brett M, Patterson K. Actions speak louder than functions: the importance of manipulability and action in tool representation. J Cogn Neurosci 2003; 15: 30–46. Kiefer M, Sim EJ, Herrnberger B, Grothe J, Hoenig K. The sound of concepts: four markers for a link between auditory and conceptual brain systems. J Neurosci 2008; 28: 12224–30. Kiefer M, Pulvermuller F. Conceptual representations in mind and brain: theoretical developments, current evidence and future directions. Cortex 2012; 48: 805–25. Lambon Ralph MA, Graham KS, Ellis AW, Hodges JR. Naming in semantic dementia—what matters? Neuropsychologia 1998; 36: 775–84. Lambon Ralph MA, Graham KS, Patterson K, Hodges JR. Is a picture worth a thousand words? Evidence from concept definitions by patients with semantic dementia. Brain Lang 1999; 70: 309–35. Lambon Ralph MA, Lowe C, Rogers TT. Neural basis of category-specific semantic deficits for living things: evidence from semantic dementia, HSVE and a neural network model. Brain 2007; 130: 1127–37. Lambon Ralph MA, McClelland JL, Patterson K, Galton CJ, Hodges JR. No right to speak? The relationship between object naming and semantic impairment: neuropsychological abstract evidence and a computational model. J Cogn Neurosci 2001; 13: 341–56. Lambon Ralph MA, Patterson K, Garrard P, Hodges JR. Semantic dementia with category specificity: a comparative case-series study. Cogn Neuropsychol 2003; 20: 307–26. Lambon Ralph MA, Sage K, Jones R, Mayberry E. Coherent concepts are computed in the anterior temporal lobes. Proc Natl Acad Sci USA 2010; 107: 2717–22. Lewis JW, Wightman FL, Brefczynski JA, Phinney RE, Binder JR, DeYoe EA. Human brain regions involved in recognizing environmental sounds. Cereb Cortex 2004; 14: 1008–21. Lezak M. Neuropsychological assessment. New York: Oxford University Press; 1976. Luzzi S, Snowden JS, Neary D, Coccia M, Provinciali L, Lambon Ralph MA. Distinct patterns of olfactory impairment in Alzheimer’s disease, semantic dementia, frontotemporal dementia, and corticobasal degeneration. Neuropsychologia 2007; 45: 1823–31. Macoir J. Is a plum a memory problem? Longitudinal study of the reversal of the conreteness effect in a patient with semantic dementia. Neuropsychologia 2009; 47: 518–35.
Brain 2012: 135; 3770–3780
| 3779
Marinkovic K, Dhond RP, Dale AM, Glessner M, Carr V, Halgren E. Spatiotemporal dynamics of modality-specific and supramodal word processing. Neuron 2003; 38: 487–97. Martin AJ. The representation of object concepts in the brain. Annu Rev Psychol 2007; 58: 25–45. Martin AJ, Haxby JV, Lalonde FM, Wiggs CL, Ungerleider LG. Discrete cortical regions associated with knowledge of color and knowledge of action. Science 1995; 270: 102–5. Mayberry EJ, Sage K, Lambon Ralph MA. At the edge of semantic space: the breakdown of coherent concepts in semantic dementia is constrained by typicality and severity but not modality. J Cogn Neurosci 2011; 23: 2240–51. Mion M, Patterson K, Acosta-Cabronero J, Pengas G, IzquierdoGarcia D, Hong YT, et al. What the left and right fusiform gyri tell us about semantic memory. Brain 2010; 133: 3256–68. Nestor PJ, Fryer TD, Hodges JR. Declarative memory impairments in Alzheimer’s disease and semantic dementia. Neuroimage 2006; 30: 1010–20. Papagno C, Capasso R, Miceli G. Reversed concreteness effect for nouns in a subject with semantic dementia. Neuropsychologia 2009; 47: 1138–48. Pexman PM, Hargreaves IS, Siakaluk PD, Bodner GE, Pope J. There are many ways to be rich: effects of three measures of semantic richness on visual word recognition. Psychon Bull Rev 2008; 15: 161–7. Patterson K, Lambon Ralph MA, Jefferies E, Woollams A, Jones R, Hodges JR, et al. “Presemantic”’ cognition in semantic dementia: six deficits in search of an explanation. J Cogn Neurosci 2006; 18: 169–83. Patterson K, Nestor PJ, Rogers TT. Where do you know what you know? The representation of semantic knowledge in the human brain. Nat Rev Neurosci 2007; 8: 976–87. Piwnica-Worms KE, Omar R, Hailstone JC, Warren JD. Flavour processing in semantic dementia. Cortex 2010; 46: 761–8. Plaut DC, Shallice T. Deep dyslexia: a case study in connectionist neuropsychology. Cogn Neuropsychol 1993; 10: 377–500. Pobric G, Jefferies E, Lambon Ralph MA. Anterior temporal lobes mediate semantic representation: mimicking semantic dementia by using rTMS in normal participants. Proc Natl Acad Sci USA 2007; 104: 20137–41. Pobric G, Jefferies E, Lambon Ralph MA. Category-specific vs. category-general semantic impairment induced by transcranial magnetic stimulation. Curr Biol 2010a; 20: 964–8. Pobric G, Jefferies E, Lambon Ralph MA. Amodal semantic representations depend on both anterior temporal lobes: evidence from repetitive transcranial magnetic stimulation. Neuropsychologia 2010b; 48: 1336–42. Pulvermuller F, Cooper-Pye E, Dine C, Hauk O, Nestor PJ, Patterson K. The word processing deficit in semantic dementia: all categories are equal, but some categories are more equal than others. J Cogn Neurosci 2010; 22: 2027–41. Raven JC. Coloured progressive matrices sets A, AB, B. Cogn Affect Behav Neurosci. London: H. K. Lewis; 1962. Rogers TT, Lambon Ralph MA, Garrard P, Bozeat S, McClelland JL, Hodges JR, et al. Structure and deterioration of semantic memory: a neuropsychological and computational investigation. Psychol Rev 2004a; 111: 205–35. Rogers TT, Lambon Ralph MA, Hodges JR, Patterson K. Natural selection: the impact of semantic impairment on lexical and object decision. Cogn Neuropsychol 2004b; 21: 331–52. Rogers TT, McClelland JL. Semantic cognition: a parallel distributed processing approach. Cambridge, MA: MIT Press; 2004. Rohrer JD, Warren JD, Modat M, Ridgway GR, Douiri A, Rossor MN, et al. Patterns of cortical thinning in the language variants of frontotemporal lobar degeneration. Neurology 2009; 72: 1562–9. Sharp DJ, Scott SK, Wise RJS. Retrieving meaning after temporal lobe infarction: the role of the basal language area. Ann Neurol 2004; 56: 836–46.
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
3780
| Brain 2012: 135; 3770–3780
Snodgrass JG, Vanderwart M. A standardized set of 260 pictures: norms for name agreement, image agreement, familiarity and visual complexity. J Exp Psychol Hum Learn Memory 1980; 6: 174–215. Thompson-Schill SL, Aguirre GK, D’Esposito M, Farah MJ. A neural basis for category and modality specificity of semantic knowledge. Neuropsychologia 1999; 37: 671–6. Tyler LK, Moss HE. Going, going, gone . . . ? Implicit and explicit tests of conceptual knowledge in a longitudinal study of semantic dementia. Neuropsychologia 1998; 36: 1313–23. Van Dam WO, van Dijk M, Bekkering H, Rueschemeyer S. Flexibility in embodied lexical-semantic representations. Hum Brain Mapp 2012; 33: 2322–33. Vandenberghe R, Price C, Wise R, Josephs O, Frackowiak RSJ. Functional anatomy of a common semantic system for words and pictures. Nature 1996; 383: 254–6. Visser M, Jefferies E, Embleton KV, Lambon Ralph MA. Both the middle temporal gyrus and the ventral anterior temporal area are crucial for multimodal semantic processing: distortion-corrected fmri evidence for a double gradient of information convergence in the temporal lobes. J Cogn Neurosci 2012; 24: 1766–78.
P. Hoffman et al. Visser M, Lambon Ralph MA. Differential contributions of bilateral ventral anterior temporal lobe and left anterior superior temporal gyrus to semantic processes. J Cogn Neurosci 2011; 23: 3121–31. Warrington EK. The selective impairment of semantic memory. Q J Exp Psychol 1975; 27: 635–57. Warrington EK, James M. The visual object and space perception battery. Bury St. Edmunds, Suffolk: Thames Valley Test Company; 1991. Warrington EK, Shallice T. Category specific semantic impairments. Brain 1984; 107: 829–54. Wechsler D. Wechsler memory scale: revised (WMS-R). New York: Psychological Corporation; 1987. Woollams AM, Cooper-Pye E, Hodges JR, Patterson K. Anomia: a doubly typical signature of semantic dementia. Neuropsychologia 2008; 46: 2503–14. Woollams AM, Lambon Ralph MA, Plaut DC, Patterson K. SD-squared: on the association between semantic dementia and surface dyslexia. Psychol Rev 2007; 114: 316–39. Yi HA, Moore P, Grossman M. Reversal of the concreteness effect for verbs in patients with semantic dementia. Neuropsychology 2007; 21: 9–19.
Downloaded from https://academic.oup.com/brain/article-abstract/135/12/3770/285941/The-degraded-concept-representation-system-in by guest on 29 September 2017
